Lecroy WR64MXi teardown...

[TunerSandwich]

I was doing some upgrades and general service on my WR64MXi, and thought you all might like to see some pics of it's guts.  I will add some closer views and further tear down the front panel, as I go through cleaning/maint and upgrades.  Enjoy. 

front panel controls and display (will add pics of what's under the hood, as I get to that section)

front end/inputs/processing (rear panel)

front end/inputs/processing (front panel)

AC/DC (main) PSU

Lecroy bus input card

Main motherboard (for OS and applications) notice the interesting daughter card, which acts as a bridge, between the MOBO and the front end main board

[mcinque]

 

What a huge PSU section!

[kizzap]

In the last picture, can you point out the "daughter-card"? The only two boards I see are the main PCB and then a generic DDR ram stick 

[free_electron]

In the last picture, can you point out the "daughter-card"? The only two boards I see are the main PCB and then a generic DDR ram stick 

put on your other glasses grampa.
it's the one in the forefront witht the bix xilinx chip on. fan left , xilinx chip then grey flatcable coming to front.
look above the xilinx chipo : see that big black pinstrip connector ? that looks like a PC104 / pci version to me

[TunerSandwich]

What a huge PSU section!

It powers all of the intermediate busses of the system...so needs to provide ATX power, as well as the bus for the input processing and front end.....it's a beast for sure.  I am a bit concerned about the density, and fact that it lives in a rather hot environment.  I can't imagine those electrolytics are going to last a decade more. 

It does have very nice discreet daughter cards and very nice input filtering.  I will show it in depth a bit more.   When it's assembled it has a nice isolated/ducted fan and shroud....

I shudder to think how much that thing is going to cost to replace when it goes.  It's also way too dense/complex to repair in house (unless I get really lucky).

[TunerSandwich]

In the last picture, can you point out the "daughter-card"? The only two boards I see are the main PCB and then a generic DDR ram stick 

[kizzap]

In the last picture, can you point out the "daughter-card"? The only two boards I see are the main PCB and then a generic DDR ram stick 

put on your other glasses grampa.
it's the one in the forefront witht the bix xilinx chip on. fan left , xilinx chip then grey flatcable coming to front.
look above the xilinx chipo : see that big black pinstrip connector ? that looks like a PC104 / pci version to me

In the last picture, can you point out the "daughter-card"? The only two boards I see are the main PCB and then a generic DDR ram stick 

http://i123.photobucket.com/albums/o282/madgreed/2014-11-25002.jpg
http://i123.photobucket.com/albums/o282/madgreed/2014-11-25001.jpg

* kizzap books an appointment to get his eyes checked.

I still think it was blended in rather well...

[jeremy]

On the rear panel of the frontend, what are those strange unmasked serpentines for? Something thermal? I imagine the grey stuff on the BGA fanout is some kind of thermally conductive tape?

[PA0PBZ]

I still think it was blended in rather well...

I agree, I had to look twice to find it. Stealth daughter card...

[Wuerstchenhund]

I am a bit concerned about the density, and fact that it lives in a rather hot environment.  I can't imagine those electrolytics are going to last a decade more. 

I wouldn't worry too much, the caps should be fine. Most of the heat is dissipated via the chassis, and the rest through the fan. The caps are also spec'd appropriately for this environment.

I was a bit shocked when I noticed that the PSU of my 64Xi is made by Cherokee, but they seem to produce pretty reliable stuff these days (they certainly haven't in the past).

I shudder to think how much that thing is going to cost to replace when it goes.

Why replace it? Just send it in to LeCroy for repair.

It's also way too dense/complex to repair in house (unless I get really lucky).

It's densely packed but I don't think it's a very complicated design. These PSUs are much more straight forward than say the dreadful Eaton PSUs in the HP 54542 scopes.

[TunerSandwich]

I am a bit concerned about the density, and fact that it lives in a rather hot environment.  I can't imagine those electrolytics are going to last a decade more. 

I wouldn't worry too much, the caps should be fine. Most of the heat is dissipated via the chassis, and the rest through the fan. The caps are also spec'd appropriately for this environment.

I was a bit shocked when I noticed that the PSU of my 64Xi is made by Cherokee, but they seem to produce pretty reliable stuff these days (they certainly haven't in the past).

I shudder to think how much that thing is going to cost to replace when it goes.

Why replace it? Just send it in to LeCroy for repair.

It's also way too dense/complex to repair in house (unless I get really lucky).

It's densely packed but I don't think it's a very complicated design. These PSUs are much more straight forward than say the dreadful Eaton PSUs in the HP 54542 scopes.

The ducting/fan system on the PSU is quite nice....I can tell it works well, by how much heat this thing kicks off 

Despite ratings and quality of those caps, I still fear their longevity....that is a pretty dense PSU, with a lot of discreet heat makers in it 

This scope is probably going to see A LOT of use....thankfully the environment is very well controlled. 

I never get lucky enough, when something fails, for it to be anything..other than the most burred surface mount BGA or QFN type device....I will gut the PSU a bit further tomorrow and have a look closer....it's certainly a bit on the sloppy side....

In general the wiring/routing in the MXi is a bit "sloppy"....it seems they could have done a lot better w/o all the delicate flying leads ETC.....the use of Kapton tape in the front panel/display area is a bit funny....either way it all comes together, just a bit surprised by how dense the entire design is.  I honestly wouldn't have minded a slightly bigger footprint, for some decreased density and better cooling.  An extra 3 inches in depth would have allowed for a much easier service access and cooling options....but it works...

[i4004]

I shudder to think how much that thing is going to cost to replace when it goes.

why would it go?

which caps in the psu, if japanese, they can last.
if they go you can put japanese caps then...

[alex.forencich]

On the rear panel of the frontend, what are those strange unmasked serpentines for? Something thermal?

Looks like high speed differential microstrip.  Probably connects the output of the front end to the ADCs.  Some people remove the solder mask over microstrip traces to get a more uniform impedance (solder mask variations will not affect trace impedance when you remove the solder mask). 

[TunerSandwich]

On the rear panel of the frontend, what are those strange unmasked serpentines for? Something thermal? I imagine the grey stuff on the BGA fanout is some kind of thermally conductive tape?

[TunerSandwich]

On the rear panel of the frontend, what are those strange unmasked serpentines for? Something thermal?

Looks like high speed differential microstrip.  Probably connects the output of the front end to the ADCs.  Some people remove the solder mask over microstrip traces to get a more uniform impedance (solder mask variations will not affect trace impedance when you remove the solder mask).

 If I have time I will remove the heatsinks on those BGA devices and we can have a look at the p/n....the traces carry over to the front board

[Lukas]

Anyone got some idea why they need these crusty looking coaxial cables all over the place? Sampling clock distribution maybe?
Are these dark things on the underside of the BGAs ferrite dampers? Interesting to see them in such place...

[T3sl4co1l]

Those are definitely microstrip, but vanishingly low impedance... trace widths like that would be reasonable on a two layer board, but...?!  I'm guessing they were going for 50 ohm differential (~25 ohm each), but geez... and why?...

It bothers me more that they didn't bother to finesse the soldermask cutouts -- it's all angular and blah!

Tim

[jeremy]

Thanks tunersandwich, this is very interesting! Would be cool to see under the heatsinks if you get a chance.

I also find it odd that the copper is not gold plated, yet it looks quite good and unoxidised. Is there lacquer on the board? Although if there was, that would probably defeat the purpose of having the bare microstrip.... Does the unit live in a humid climate?

edit: also, it just occurred to me that they are in fact not length matched....

[T3sl4co1l]

I've got at least another board with unplated copper,





(Note, second pic is at a right angle to the first; note the bigger and smaller empty holes, and the common hole with a brass insert.)

This is a HP/Compaq 8510W motherboard; this area is between the Northbridge and Super IO chips.

I would assume they used a somewhat more reactive flux in the solder paste, relative to the recommended low activity flux, RoHS paste and gold plated boards, that are most common for assemblies.

The copper parts look coppery, but remember, a bright copper surface is... well, *bright*!  Almost as bright as the brass.  So there's more than a few years of tarnish on there (almost five, in this case, actually).

Tim

[TunerSandwich]

Anyone got some idea why they need these crusty looking coaxial cables all over the place? Sampling clock distribution maybe?
Are these dark things on the underside of the BGAs ferrite dampers? Interesting to see them in such place...

That coax looks pretty far from crusty.  It's well terminated and routed.  I will trace them and try to get an idea of what they are doing....I have a feeling something to do with ext trig and other front end duties, but I haven't really investigate.  Will have a look...I am curious now that you brought it up.

The pads on the back of the main BGA in the cluster, are silicon and have a rough texture....seems like isolation, rather than thermal.....I will lay it in the chassis and see what they might line up with.....

the two BGA clusters are obviously driving or processing channel pairs.  1-2 + 3-4.....the main BGA cluster with cooling fan has to be post a/d and acquisition.....I wouldn't be surprised if that BGA clusters traces lead back to the ribbon cable connections, that go to the PCi bridge board.....I will beep those out and see where they go.....

[marshallh]

Any pictures of the acquisition section?

[TunerSandwich]

Those are definitely microstrip, but vanishingly low impedance... trace widths like that would be reasonable on a two layer board, but...?!  I'm guessing they were going for 50 ohm differential (~25 ohm each), but geez... and why?...

It bothers me more that they didn't bother to finesse the soldermask cutouts -- it's all angular and blah!

Tim

They carry over to the front of the board and disappear under the passive heat sinks.  Let's lift those and remove the i/o board and see....

[TunerSandwich]

Any pictures of the acquisition section?

specifically the i/o?

[Wuerstchenhund]

That coax looks pretty far from crusty.  It's well terminated and routed.  I will trace them and try to get an idea of what they are doing....I have a feeling something to do with ext trig and other front end duties, but I haven't really investigate.  Will have a look...I am curious now that you brought it up.

If I remember right the coax is for the internal calibration signals for the 50ohms and 1Mohms paths. This is used when the scope stops acquisition and goes into calibration cycle to compensate for temperature variations and changes in the signal path.

On these scopes there's a setting to disable internal calibration for thermal compensation but will still re-calibrate when larger changes in the vertical setting are made.

[TunerSandwich]

Front panel USB plus SigGen/Test/Cal out



CH input topside...notice the interesting resistive divider at the top left?



CH inputs topside



CH inputs bottom side



There is where our "crusty" coax leads to/from....it terminates on more of the same traces as the backside of the board...the traces run up the board under the cans, through routed channels....3 cans, 3 coax.....not about to bother lifting the cans off....but I think the auto cal circuitry is the best bet for those sections



Topside of processing board, with the inputs/acquisition board removed....love the hirose board to board connections....will possibly pull those heat sinks off tomorrow and expose the BGA clusters

[Wuerstchenhund]

Topside of processing board, with the inputs/acquisition board removed....love the hirose board to board connections

Those connectors are actually the weak spot on the WaveRunner (M)Xi Series. as they tend to loose contact due to thermal stress. As a result, one or more channels (usually its Ch 2 and 3) won't show a signal (or only a very distorted one), and when using the vertical offset for those channels the trace moves as long as the know is turned but as soon as the adjustment stops the trace jumps back to somewhere near the center.

LeCroy does repair the connectors, although at a price. If they fail, maybe they can be replaced with a more durable solution.

[TunerSandwich]

Topside of processing board, with the inputs/acquisition board removed....love the hirose board to board connections

Those connectors are actually the weak spot on the WaveRunner (M)Xi Series. as they tend to loose contact due to thermal stress. As a result, one or more channels (usually its Ch 2 and 3) won't show a signal (or only a very distorted one), and when using the vertical offset for those channels the trace moves as long as the know is turned but as soon as the adjustment stops the trace jumps back to somewhere near the center.

LeCroy does repair the connectors, although at a price. If they fail, maybe they can be replaced with a more durable solution.

Not sure why those would lose contact.....quite tight and bolted and braced.....I have used similar Hirose b2b connectors in quite a few places, including high thermal, high stress applications, and never once had a problem.  If they are losing contact/continuity it has to be on the board side, not the b2b joint. 

In fact I am using them on a daughter card (high current) POL...in a battery powered portable design.  The devices sitting on the card are pushing over 15 amps and run consistently at around 74deg C....their passive cooling solution dissipates straight into the PCB.....not one failure in testing or use.  The connectors don't carry the high current up to the card, but they carry critical and sensitive serial (bit corn) and other command/protocol data....those signals are free of any degradation, even beyond nominal operation temps. 

Is the issue documented at all?  If so what was the resolution?  rework of the pad to edge? 

[Wuerstchenhund]

Not sure why those would lose contact.....quite tight and bolted and braced

Exactly, which is the cause of the problem.

In fact I am using them on a daughter card (high current) POL...in a battery powered portable design.  The devices sitting on the card are pushing over 15 amps and run consistently at around 74deg C....their passive cooling solution dissipates straight into the PBC.....not one failure in testing or use.

I don't know your design of course but it may well be that the PCB has plenty of room to expand. In the WR(M)Xi, the fixing posts prevent the aquisition board from having space to expand when heating up, which causes thermal stress which results in stress on the connectors themselves.

Is the issue documented at all?  If so what was the resolution?  rework of the pad to edge?

My scope is one of the affected ones, and according to LeCroy support this is a well known problem with the (M)Xi. Not sure if its documented outside but that is probably true for most things of these newer scopes.

The solution they offer is a board repair, apparently involving replacing the connectors. It's not cheap, but of course still a lot less expensive than just swapping the boards.

[TunerSandwich]

That is very interesting....

When we use Hirose b2b connectors we always use a small elastomer or fiber washer, on the standoffs, that bolt the daughter cards to the main boards.  Hirose themselves recommend this "finger tight" fit.

I wonder if simply putting a thin fiber washer at the base of those bracing blocks, and not going more than "finger tight" would help this issue.  It would allow for about 0.15mm expansion...

LeCroy is doing virtually the same thing on the Hirose connector on the back of the main board, where the probus interface card mounts.  They have a longer screw, a thin belville washer, and a nylon spacer on the back of the screw. 

Another option would be a small sliver of silpad to "float" the acquisition board a bit off the main board. 

Will try both of these methods and see if it causes any issues.

[Wuerstchenhund]

Since you have a later production date it might well be that your scope already has provisions for alleviating the problem.

My unit from 2005 unfortunately has been hit (Ch 2 and 3 are affected). I also don't have the tools to re-solder or replace the connectors. Luckily I still have two other scopes.

BTW: if anyone knows a shop in the UK or in Europe who can replace the connector and does a thorough job then please let me know, as the LeCroy repair is quite expensive.

[Lukas]

This is used when the scope stops acquisition and goes into calibration cycle to compensate for temperature variations and changes in the signal path.

On these scopes there's a setting to disable internal calibration for thermal compensation but will still re-calibrate when larger changes in the vertical setting are made.

I've always been wondering why LeCroy scopes need to do this, as the rest of the industry gets away without interrupting the user every now an then...

[TunerSandwich]

Since you have a later production date it might well be that your scope already has provisions for alleviating the problem.

My unit from 2005 unfortunately has been hit (Ch 2 and 3 are affected). I also don't have the tools to re-solder or replace the connectors. Luckily I still have two other scopes.

BTW: if anyone knows a shop in the UK or in Europe who can replace the connector and does a thorough job then please let me know, as the LeCroy repair is quite expensive.

Do you possibly have a picture of the failed pads?  VERY curious to see exactly how it fails.  I am going to stick the fiber washers in mine, as a precaution. 

From what I can tell on my unit, most of the chips and the lecroy code says around Sept. 2009....but the HDD lot code is 2005.....

BTW my fans are Young Lin, rated 4.2W.  The blade profile is one of an air mover.....def not a static pressure fan.  Does any of that ring a bell, or are these the most current fan rev?  Still considering swapping them for some Delta air movers I have (wish I knew how much total current the fan regulators are rated for).

[Wuerstchenhund]

Do you possibly have a picture of the failed pads?  VERY curious to see exactly how it fails.

Unfortunately not. I had it open (for the upgrades) but then put it back together. I didn't look closer to what's wrong with the connectors, but my gut feeling is that maybe they just need resoldering. But as I lack the tools I just put it together and now occasionally use it with two channels only. If I can't find a cheaper alternative then I'll have to send it to LeCroy for repair.

From what I can tell on my unit, most of the chips and the lecroy code says around Sept. 2009....but the HDD lot code is 2005.....

I wouldn't worry about the hard disk, maybe it has been replaced at some points in the past or they bought a large charge in 2005 (which wouldn't be surprising, considering that by then SATA was the newest thing).

BTW my fans are Young Lin, rated 4.2W.  The blade profile is one of an air mover.....def not a static pressure fan.  Does any of that ring a bell, or are these the most current fan rev?

Sounds like your scope already has the uprated fans.

Still considering swapping them for some Delta air movers I have (wish I knew how much total current the fan regulators are rated for).

I'd just leave it as it is.

[TunerSandwich]

I was able to successfully "float" the acquisition board, from the main processing board.  I tried nylon and fiber washers, but then realized the through holes (bolt down points) were part of the ground/isolation plane (beeped them out).  So I wound up using split lock washers (very small aluminum) on the backside of the main board, where the screws feed through.  I tested the expansion and it offers a decent amount of "play" in the connector.  enough to allow for movement under a thermal condition. 

Everything fit fine in the chassis, and the backspacing on the screws doesn't interrupt any critical mechanical spacing.

Scope is back together and running flawlessly.  I also modified the spacing on the LC bus daughter card, using a nylon washer on top (replacing the steel unit). 

I did some testing on the draw of the delta fans, and they weren't acceptable (over 10 watts/fan). 

With the new CPU the cpu fan is running on overdrive 90% of the time.  I used some Antec series 7 thermal compound on the CPU block, but it just isn't a large enough mass or fan to cope with the increased heat.  I am going to have a new CPU block cut and utilize some more of the horizontal space in the mobo enclosure.  I noticed a chassis opening just beside the block, so I might have it cut and install a semi-external passive cooling solution.  The amount of heat being kicked out of the scope is WAY beyond where it was with the Celeron.  I am a bit concerned, especially since the CPU block is adjacent to a section of the PSU, which contains a lot of electrolytic caps.  Overall lecroy didn't do very well (IMO) as far as cooling solutions for this chassis. 

This might all be un-necessary/overkill, but if anything the added noise from the CPU fan running at full tilt all the time is enough to justify the attempt (for me). 

When I strip it down again, for HDD and wiring changes, I will try these mods.  I found a slick, straight through, IDE/SATA bridge....I was going to modify it's connectors and mount, to clamp directly to the replacement SATA drives chassis.  I want to minimize the ribbon cable slack and keep it from bunching in the chassis (obsessive I know).

i will update this thread with more pictures of the mods, when I tear it down again.  I might also modify the MOBO chassis upper panel (that holds the LC bus connector) to space a tad bit higher.  It doesn't align as well as it should, and I am concerned about potential strain on the connector.  A couple slim nylon washers should give it a extra mm and bring the connector up just a bit on the side panel cutout.

[Wuerstchenhund]

With the new CPU the cpu fan is running on overdrive 90% of the time.  I used some Antec series 7 thermal compound on the CPU block, but it just isn't a large enough mass or fan to cope with the increased heat.

The heat sink and fan should be sufficient for your processor. However, unlike the old Celeron, the Pentium-M does support SpeedStep which clocks the CPU down if the load is low. XP supports SpeedStep, but depending on the power settings it might just nail the CPU down on constant full throttle.

I think (can't check as I'm at work at the moment) that SpeedStep also needs to be enabled in the BIOS. The same might be true for temperature control of the fan.

I am going to have a new CPU block cut and utilize some more of the horizontal space in the mobo enclosure.  I noticed a chassis opening just beside the block, so I might have it cut and install a semi-external passive cooling solution.  The amount of heat being kicked out of the scope is WAY beyond where it was with the Celeron.  I am a bit concerned, especially since the CPU block is adjacent to a section of the PSU, which contains a lot of electrolytic caps.  Overall lecroy didn't do very well (IMO) as far as cooling solutions for this chassis.

Well, its narrowly built, but don't forget that this design has proven to be running fine in the field for now more than 9 years with no heat issues (aside from the acq board connector problem on earlier batches) or higher than normal PSU failures.

The other thing is that if you modify too much (i.e. more than just RAM, CPU and hard disk) LeCroy may refuse repair if you ever need it.

However, I agree that the build quality is only average on the (M)Xi Series, and no match for the Iwatsu-made scopes (WR, WR2, WP900) or the first and second generations of X-Stream scopes (WM8k, WP7k, WR6k).

When I strip it down again, for HDD and wiring changes, I will try these mods.  I found a slick, straight through, IDE/SATA bridge....I was going to modify it's connectors and mount, to clamp directly to the replacement SATA drives chassis.  I want to minimize the ribbon cable slack and keep it from bunching in the chassis (obsessive I know).

As before, I wouldn't recommend to go further with modifications to such a scope, especially when every mod can easily introduce new problems. You may end up with an unreliable piece of kit (I understand you bought it for your business, so I guess you want something reliable).

[TunerSandwich]

I will dig around the BIOS and OS tonight.  I was trying to avoid going in and modifying core system parameters (until I get that drive backed up). 

I am also NOT noticing a real improvement in scope application processing (maybe it's me but it honestly seems worse than before).  However the boot times are VERY much improved.

I will post some screenshots....when I run SpecAn the CPU throttles up to max load and pegs there, while the UI completely locks out.....something seems a bit wrong about that

[Wuerstchenhund]

I will dig around the BIOS and OS tonight.  I was trying to avoid going in and modifying core system parameters (until I get that drive backed up).

There's not much to modify in Windows. Just go to Control Panel - Power settings and change the power profile from the high power profile ("always on" or something similar, can't remember at the moment) to a flexible profile (i.e. "Laptop").

I am also NOT noticing a real improvement in scope application processing (maybe it's me but it honestly seems worse than before).

That's strange, as the difference should be quite noticable. It does suggest that there's something wrong with your setup.

Since you say it's worse than before, and your system seems to be running excessively hot, it may well be that the CPU runs into throtteling due to overheating, which would explain why it's so slow.

I can switch on mine and check the temperature readings tonight.

I will post some screenshots....when I run SpecAn the CPU throttles up to max load and pegs there, while the UI completely locks out.....something seems a bit wrong about that

Indeed, this shouldn't happen. The old Celeron-M 310 1.3GHz has a TDP of 24.5W and lacks SpeedStep, so it's running full throttle constantly. The P-M 745 1.8GHz I have in my scope has only 21W, and supports SpeedStep so won't run full throttle unless the load is high (and even then puts out less heat). Even the fastest P-M processors only put out 27W which is only 2.5W more than the Celeron M, and again they run with SpeedStep. You shouldn't really see any increase in heat after the CPU swap.

[TunerSandwich]

I was playing around in BIOS earlier today and found a problem.....the temp throttling was set at 60 deg C and the CPU was idling at 61 deg C.....playing around with some settings now....also need to play with some FSB speed settings to match up this RAM and going to manually set timings.....I think the SPD on this stick of ram is wonky.....the timings are off of the manufacturer stated ratings....

I normally would just set all of this stuff manually and move along, but I am a bit paranoid...seeing as I have a real lack of experience with LeCroy scopes.....I am a bit paranoid to just treat it like "another computer"....probably being over cautious.....baby steps 

[abraxa]

Unfortunately, the pictures are no longer available. Would it be possible for you to upload them elsewhere and replace the links if it's not too much work?